Speed scaling to manage temperature
STACS'05 Proceedings of the 22nd annual conference on Theoretical Aspects of Computer Science
| Hi-index | 0.00 |
In our research, we study how real-time systems are affected by thermal management to satisfy the temperature constraint. In temperature-constrained real-time systems, deadline guarantees must be met without exceeding safe temperature level of the processor. While processor speed control is the most popular method of thermal management of systems, it eventually makes the task delays longer. In our study, we describe how to find the worst case execution considering speed control in temperature-constrained environment. With the worst case execution scenario, we study how the simple reactive speed scaling scheme can improve the processor utilization compared with any constant-speed scheme. For aperiodic tasks, it is briefly reviewed how the naive application of slack stealing leads to missed deadlines and the design-time slack allocation is proposed. A queueing model is presented to analyze the response time provided to aperiodic jobs and validated with results from a discrete-event simulator.